This is the source for @godtributes, a Twitter bot that posts tweets in the format:
SOMETHING FOR THE SOMETHING GOD
It will also occasionally reply to followers in the same way, using words they've said.
It runs on Node and uses the excellent twit module to interact with Twitter.
Warning: For now, it depends on the stable branch of level-cache-server running on the machine and will have problems with master.
git clone git@github.com:jimkang/godtributes.git
npm install
Then, create a config.js file in the project root that contains your Twitter API keys and your Wordnik API key. Example:
module.exports = {
twitter: {
consumer_key: 'asdfkljqwerjasdfalpsdfjas',
consumer_secret: 'asdfasdjfbkjqwhbefubvskjhfbgasdjfhgaksjdhfgaksdxvc',
access_token: '9999999999-zxcvkljhpoiuqwerkjhmnb,mnzxcvasdklfhwer',
access_token_secret: 'opoijkljsadfbzxcnvkmokwertlknfgmoskdfgossodrh'
},
wordnikAPIKey: 'mkomniojnnuibiybvuytvutrctrxezewarewetxyfcftvuhbg'
};
Set up an instance of level-cache-server in a parallel directory.
Optionally, set up a cron job for run maketributes.js and/or exhort.js periodically.
To post a tribute ([something] FOR THE [something] GOD) tweet, run the commmand:
node maketribute.js
To make a tribute and print it to the console, but not post it:
node maketribute.js --simulate
To make tribute replies to followers:
node exhort.js
To make tributes for followers and print them to the console, but not post them:
node exhort.js --simulate
godtributes now uses iscool's blacklists. If you'd like to add to them, make an issue or PR there.
Run tests with make test.
maketribute.js is the script that makes tributes. It uses wordniksource to get topics for the tributes, figurepicker for the figures of the tributes, and tributedemander to construct the tributes.
wordniksource is an adapter for the Wordnik API. It makes calls to Wordnik to get random words, find word frequencies, and get parts of speech.
tributedemander puts together the "[something] FOR THE [something]" strings. It uses canonicalizer to get the proper forms of the topic and figure words involved.
canonicalizer figures out if a word is singular, plural, or possessive (largely using the inflection module), then generates non-possessive singular and plural forms.
figurepicker rolls on a RPG-style table to generate figures for tribute. The main tribute figure has a 75% chance of being 'GOD'.
exhort.js is the script that gets the followers of @godtributes, decides whether to tweet at them, then comes up with a tribute to tweet at them. It uses all of the above modules, plus nounfinder to find interesting nouns within a follower's tweet and chronicler to keep track of what it's already done.
nounfinder uses a few internal rules and wordniksource for parts of speech and frequency to decide what words are interesting nouns.
chronicler uses a LevelDB database (via basicset-levelwrap) to track which tweets have been replied to and when a user was last replied to.
MIT.